VOLUME 81, NUMBER 26 P H Y S I C A L R E V I E W L E T T E R S 28 DECEMBER 1998

5804
Demonstration of a High Average Power Tabletop Soft X-Ray Laser

B. R. Benware, C. D. Macchietto, C. H. Moreno, and J. J. Rocca
Department of Electrical Engineering, Colorado State University, Fort Collins, Colorado 80523

(Received 31 August 1998; revised manuscript received 9 November 1998)

We report the first demonstration of a high average power tabletop soft x-ray laser. An average
laser output power of ø1 mW s.2 3 1014 photonsysd was generated at 46.9 nm in Ne-like Ar using a
very compact tabletop discharge. The spatially coherent average power emitted by this 26.5 eV laser is
comparable to that generated at this photon energy in a similar bandwidth sDlyl ­ 1024d by a third-
generation synchrotron beam line. Lasing was obtained at a repetition rate of 7 Hz with an average
output energy of 135 mJypulse by exciting a plasma column in a ceramic capillary with a fast current
pulse. This very compact high-repetition-rate laser source makes intense short-wavelength coherent
radiation accessible to a wide variety of new applications. [S0031-9007(98)08022-3]

PACS numbers: 42.55.Vc, 52.55.Ez, 52.75.Va
Important motivations for the development of high av-
erage power tabletop sources of coherent soft x-ray radia-
tion can be found in applications belonging to several
disciplines of science and technology. To date, the
leading sources of high average power coherent soft x-ray
radiation have been undulators in modern storage rings,
so-called third-generation synchrotrons, in which ra-
diation is emitted by a highly relativistic electron
beam as it traverses a periodic magnetic structure [1].
Synchrotrons are used successfully in many areas in-
cluding material surface diagnostics, photochemistry and
photophysics, atomic and molecular physics, very high-
resolution metrology, and biology. They have the
important advantages of having a high average brightness
and a broad tunability. However, they are typically large
facilities that are very expensive to construct and operate.

High average power beams of soft x-ray coherent
radiation can in principle also be generated with much
more compact devices using either nonlinear optical
techniques for frequency up-conversion of optical laser
radiation or the direct amplification of spontaneous
emission in a plasma (an x-ray laser). At present, the
generation of high order harmonics under fairly opti-
mized nonphased matched conditions typically yields a
conversion efficiency of about 1026 in the photon energy
range of 10–40 eV (of the order of 109 photons per
pulse) [2]. The highest pulse energy generated in a high
order harmonic pulse to date is 60 nJ at a photon energy
of about 50 eV, and corresponds to an experiment done
using the second harmonic of a powerful Nd-glass laser
system [3]. Recently, a very important advance in this
field was accomplished with the first demonstration of
phase-matched harmonic conversion of visible light into
soft x rays [4]. A conversion efficiency of 1025 1026

was obtained in the 40–70 eV spectral region. Soft
x-ray pulses with an energy of .0.2 nJ per harmonic
order were produced at a repetition frequency of 1 kHz,
corresponding to an average power of .0.2 mW [4].
This result is significant because further optimization
of this technique could result in the generation of high
0031-9007y98y81(26)y5804(4)$15.00
average power beams of spatially coherent soft x-ray
radiation from a tabletop optical laser.

Alternatively, soft x-ray lasers have the advantage of a
much higher energy per pulse and therefore the potential
to produce soft x-ray beams with very high average
powers. X-ray lasers pumped by large optical lasers of
the type used in fusion research have generated output
pulse energies of up to several mJ [5]. However, these
lasers are limited to a repetition frequency of only a few
pulses per hour at best. Recently, significant progress
has been accomplished in the development of more
compact soft x-ray lasers that can operate at increased
repetition frequencies [6–11]. Soft x-ray lasers based
on the transient collisional excitation scheme, which are
pumped by picosecond lasers that typically occupy a few
optical tables, have produced soft x-ray amplification at
repetition rates up to one shot every 3 min [6,7]. Our
group has previously reported the generation of laser
output pulse energies up to 25 mJ in the J ­ 0-1 line of
Ne-like Ar at 46.9 nm from single pass amplification in a
compact capillary discharge tabletop soft x-ray amplifier
[11]. However, in all these lasers the average power
obtained to date has been very low, as a result of the low
repetition rate (# 2 shots per min). In contrast, soft x-ray
amplification at high repetition rates has been obtained
utilizing relatively compact high intensity ultrashort pulse
lasers to pump soft x-ray lasers in plasmas created by
optical-field-induced ionization [8,9]. Amplification was
demonstrated at 10 Hz at 41.8 nm in collisionally excited
Pd-like Xe (11 to 12 gain-lengths) [8], and at 2 Hz at
13.5 nm in H-like Li following plasma recombination
(ø5.5 gain-lengths) [9]. Similar gain was obtained at
1 Hz at 26.2 nm in H-like B in a recombining plasma
excited by the combination of two low power laser pulses
[10]. However, the average power obtained in these laser-
pumped soft x-ray amplifier systems was small due to
the low laser output pulse energy, and required the use
of high-gain microchannel plates to detect the laser line.
An important next step in the development of practical
tabletop soft x-ray lasers is the generation of high average
© 1998 The American Physical Society



VOLUME 81, NUMBER 26 P H Y S I C A L R E V I E W L E T T E R S 28 DECEMBER 1998
output powers, which simultaneously requires a high
repetition rate and a high energy per pulse. In this Letter
we report the first demonstration of a high average power
tabletop soft x-ray laser. An average power of 0.95 mW
was generated at 46.9 nm. This value exceeds the average
powers obtained to date with tabletop soft x-ray lasers by
2 to 3 orders of magnitude. Our results were obtained by
generating an axially uniform plasma column with a fast
current pulse at a repetition rate of 7 Hz in an alumina
sAl2O3d capillary filled with preionized Ar gas. The
capillary discharge-pumped laser used in this experiment
occupies an area of approximately 0.4 m 3 1 m on top of
a table, a size comparable to that of many widely utilized
visible or ultraviolet gas lasers.

In capillary discharges the amount of material ablated
from the walls by plasma radiation and electron heat
conduction can greatly affect the plasma compression
and heating. In the polyacetal capillaries utilized in our
previous discharge-pumped collisional soft x-ray laser
experiments 20% to 50% of the ø39 kA discharge current
necessary to excite Ne-like Ar is computed to flow
through material ablated from the walls [12]. This signifi-
cantly reduces the efficiency of the plasma compression.
In the present experiment, we utilized Al2O3 capillaries
that are much more resistant to ablation and also have
a larger heat conductivity that allows for the dissipation
of high average discharge powers. An indication of the
dramatic reduction in the mass of ablated material in
the ceramic capillaries as compared to the polyacetal
capillaries is the much smaller pressure increase measured
at the exit of the capillary channel following a discharge
shot: ,10 mTorr instead of several Torr. This reduced
wall ablation in ceramic capillaries results in a more
efficient use of the current pulse, which allows for a
significantly lower excitation current and for operation
at high repetition rates. Lasing with an average power
of 0.95 mW was obtained by exciting Ar filled alumina
capillaries 3.2 mm in diameter and 18.2 cm in length,
with a current pulse having an amplitude of ø24 kA,
a 10% to 90% rise time of ø25 ns and a first half-
cycle duration of ø110 ns. The discharge setup and
pulse generator used in this experiment resemble those we
described previously [13,14]. The fast current pulse was
produced by discharging a water capacitor through a spark
gap switch connected in series with the capillary load.
The water served as a liquid dielectric for the capacitor
and also cooled the capillary. The capacitor was pulse
charged by a four-stage Marx generator that is enclosed
in a separate box and connected to the laser head with a
coaxial cable [11]. A nearly optimum Ar gas pressure
of 490 mTorr was maintained in the capillary using a
continuous Ar flow and a differential pumping system.

To determine the average output power the laser
output pulse energy was measured for every shot using
a vacuum photodiode placed at 87 cm from the exit of
the laser. The data were recorded and stored by a 2 3
109 samplesysec digitizing oscilloscope with 500 MHz
analog bandwidth. The quantum efficiency of the Al
photocathode was previously calibrated with respect to a
silicon photodiode of known quantum yield [12]. The
laser output was attenuated with several stainless steel
meshes of measured transmissivity to avoid saturation of
the photodiode. We have successfully operated the laser

FIG. 1. Measured output pulse energy and average output
power of the 46.9 nm laser corresponding to 1 min of con-
tinuous operation at a repetition rate of 7 Hz. (a) Shot to shot
laser output pulse energy. (b) Average output power computed
as a running average of 50 contiguous laser pulses. (c) Distri-
bution of the output pulse energy. The average pulse energy is
135 mJ and the standard deviation is 17 mJ.
5805



VOLUME 81, NUMBER 26 P H Y S I C A L R E V I E W L E T T E R S 28 DECEMBER 1998
at repetition frequencies up to 10 Hz. However, the most
detailed optimization was performed at 7 Hz, resulting
in the highest average power at this lower frequency.
Figure 1 shows the measured laser output pulse energy
and average power at 7 Hz repetition frequency at the
near-optimum discharge pressure of 490 mTorr. The data
correspond to 1 min of continuous operation of the laser.
Figures 1(a) and 1(c) show that the average output energy
per pulse is s135 6 17d mJ, and Fig. 1(b) shows that the
average power at 7 Hz is 0.95 mW. This corresponds to
.2.2 3 1014 photons per second. We have operated the
laser uninterruptedly during 30 min at 5 Hz, and we have
obtained up to 9 3 103 laser shots from a single capillary.
After this number of shots lasing was still observed,
but the output pulse energy degraded to about half its
maximum value due to deterioration of the capillary walls.

The laser pulse width was measured using a vacuum
photodiode and an analog oscilloscope with 1 GHz band-
width. The measurements were corrected by taking into
account the limited frequency response of the oscillo-
scope. Figure 2 shows the temporal profile of a typi-
cal laser pulse. The full width at half maximum pulse
width of the laser pulses generated at 7 Hz was mea-
sured to be ø1.2 ns. The laser pulse width was mea-
sured to be slightly larger at 1 Hz, ø1.3 ns. The small
reduction of the laser pulse width with increased repeti-
tion rate is most likely associated with a reduction of the
density of Ar atoms in the capillary channel as a result
of increased gas heating. This is supported by measure-
ments that showed a decreased laser pulse width at re-
duced Ar pressures. The peak power of a typical laser
pulse is ø112 kW, while the peak power of the most in-
tense shots exceeds 150 kW. The laser beam divergence
and spatial intensity distribution were measured with a
two-dimensional x-ray sensitive detector placed at 138 cm
from the exit of the capillary. The detector consisted of a

FIG. 2. Temporal evolution of the laser intensity for a typical
laser shot generated at a repetition frequency of 7 Hz. The
solid curve is the photodiode signal recorded with a 1-GHz-
bandwidth analog oscilloscope. The dashed curve is the signal
corrected for the limited bandwidth of the oscilloscope.
5806
phosphor screen and a charge-coupled device (CCD) array
of 1024 3 1024 pixels. Figure 3 illustrates the measured
spatial intensity distribution of the beam that has an annu-
lar shape. The annular shape is the result of refraction of
the amplified rays in the plasma column due to a plasma
density gradient in the radial direction [15]. The peak to
peak beam divergence is ø4 mrad.

It is of interest to compare the spatially coherent
power emitted by this tabletop laser to that generated
at 26.5 eV in a similar bandwidth by a third-generation
synchrotron beam line. For the purpose of making a
comparison of output power per unit bandwidth, the
laser linewidth can be conservatively estimated to be
Dlyl ­ 1 3 1024. This results from considering that
the Doppler broadened linewidth of the laser transition
for an ion temperature of Ti ø 100 eV [12] is about
Dlyl ­ 1.2 3 1024, and that it narrows by a factor of
sgplsatd1y2 ø 3.7 as the radiation is amplified to reach
saturation at gplsat ø 14 [12]. While some degree of
line rebroadening might occur after saturation [16], the
laser linewidth can be expected to still remain below
Dlyl ­ 1 3 1024. Assuming that the spatial coherence
of the laser beam produced in the ceramic capillary is
similar to that recently measured for a discharge in a
polyacetal capillary (about 6 times diffraction limited in
the tangential direction and 8 to 9 times diffraction limited
in the radial direction [17]), the average spatially coherent
power can be estimated to be about 2% of the 0.95 mW
measured, or ø19 mW. This spatially coherent power is
similar to the ø20 mW emitted at Dlyl ­ 1 3 1024 by
the 8 cm period undulator at the Advanced Light Source,

FIG. 3. Image of the laser beam at 1.38 m from the exit of
the capillary. The intensity distribution corresponds to the sum
of five consecutive shots at 7 Hz. The annular beam pattern is
caused by refraction.



VOLUME 81, NUMBER 26 P H Y S I C A L R E V I E W L E T T E R S 28 DECEMBER 1998
a third-generation synchrotron situated at Berkeley [18].
Nevertheless, for some applications the synchrotron has
the advantage of broad tunability at shorter wavelengths.
In contrast, the peak power of the spatially coherent
radiation emitted by the capillary discharge laser is
estimated to be about 2.3 kW, and exceeds by more than
5 orders of magnitude that of the undulator, estimated at
ø4 mW.

In conclusion, we have demonstrated the operation of
a very compact tabletop 46.9 nm laser at an average
power of ø1 mW. This result corresponds to an increase
in the average power available from soft x-ray lasers
by 2 to 3 orders of magnitude. The demonstration
of a tabletop soft x-ray laser with a spatially coherent
average power per unit bandwidth similar to that of a
synchrotron beam line has the potential to greatly expand
the use of intense coherent short-wavelength radiation in
applications.

We thank Vyacheslav Shlyaptsev and Siu Au Lee
for many useful discussions and Elliot Bernstein for his
interest and encouragement. We also acknowledge the
assistance of Bill McHardy. This work was supported by
the National Science Foundation under Grant No. DMR-
9512282.
[1] D. T. Attwood, K. Halbach, and N. J. Kim, Science 228,
1264 (1985).

[2] A. L’Huillier, in X-Ray Lasers 1996, edited by S.
Svanberg and C. G. Wahlstrom (Institute of Physics,
Bristol, 1996), p. 444.

[3] T. Ditmire et al., Phys. Rev. A 51, R902 (1995).
[4] A. Rundquist et al., Science 280, 1412 (1998).
[5] L. B. DaSilva et al., Opt. Lett. 18, 1174 (1993).
[6] P. V. Nickles et al., Phys. Rev. Lett. 78, 2748 (1997).
[7] J. Dunn et al., Phys. Rev. Lett. 80, 2825 (1998).
[8] B. E. Lemoff et al., Phys. Rev. Lett. 74, 1576 (1995).
[9] D. Korobkin, C. H. Nam, S. Suckewer, and A. Golstov,

Phys. Rev. Lett. 77, 5206 (1996).
[10] D. Korobkin, A. Goltsov, A. Morozov, and S. Suckewer,

Phys. Rev. Lett. 81, 1607 (1998).
[11] B. R. Benware, C. H. Moreno, D. J. Burd, and J. J. Rocca,

Opt. Lett. 22, 796 (1997).
[12] J. J. Rocca et al., Phys. Rev. Lett. 73, 2192 (1994).
[13] J. J. Rocca, D. P. Clark, J. L. A. Chilla, and V. N. Shlyapt-

sev, Phys. Rev. Lett. 77, 1476 (1996).
[14] J. J. Rocca et al., Phys. Plasmas 2, 2547 (1995).
[15] C. H. Moreno et al., Phys. Rev. A 58, 1509 (1998).
[16] J. Koch et al., Phys. Rev. Lett. 68, 3291 (1992).
[17] M. C. Marconi et al., Phys. Rev. Lett. 79, 2799 (1997).
[18] D. T. Attwood et al., IEEE J. Quantum Electron. (to be

published) (see Fig. 8, adjusted for Dlyl ­ 1024).
5807